[Cosmo-dev] 0.6 sharing proposal

[Morgen Sagen]

On Nov 1, 2006, at 4:27 PM, Brian Moseley wrote:

> On 11/1/06, Morgen Sagen <morgen at osafoundation.org> wrote:
>
>> I'm definitely willing to combine any of these steps as needed; it
>> seems like Randy is suggesting combining phases 1 and 3, followed up
>> by phase 2.  Brian is suggesting combining *all* of them into one
>> phase I think.  I need to think about this some more before I make my
>> vote, but I just wanted to let you know I am reading this. :-)
>
> i think randy's proposal is more like phase 1 + half of phase 3. he's
> not suggesting a single http request for a sync or publish.

Oh, I thought he was; from Randy's wiki page:

"There will be a single PUT of all the EIM records that make up the  
shared collection, and then a PUT for each .ics resource.  
Synchronizing the collection will require a single sync call to get  
changes on the server using the new sharing protocol as well as a  
sync for calendar events."

>
> to be totally honest, i don't think chandler has enough release cycles
> left before preview to break the process into three phases. i can be
> convinced otherwise, but history seems to be on my side.

I didn't mean to imply a one-to-one correspondence between "phase"  
and Chandler alpha release. The three phases simply represent my plan  
of attack, each shorter than a release cycle.  In fact my phase 2 is  
simply turning off a bunch of Chandler code.  :-)  Phase 1 --  
actually implementing EIM record import/export -- will likely be the  
bulk of my alpha5 time.  Phase 3 is a bonus if we get that feature  
from Cosmo.

> Phase 1 would have Chandler continue to publish .ics files to a  
> calendar collection on Cosmo, with a ".chandler" subcollection  
> containing .xml resources.  The only difference from today would be  
> that the .xml files would contain EIM records instead of the  
> current XML Chandler uses.
>
> Phase 2 would take place when Cosmo started understanding those EIM  
> records, and Chandler could stop publishing the .ics files. (No  
> more dual-fork headaches!)
>
> Phase 3 would do away with DAV and use a single HTTP operation to  
> GET or POST a bunch of changes to a collection all at once.  (ACID  
> transactions, baby!)